Skip to main content

Full text of "Annual Evaluation Report Fiscal Year 1984"

See other formats


101-7

C. Costs, Benefits, and Effectiveness; (continued)

Table 4 reports achievement data from all States except Connecticut* the
District of Columbia, Michigan, New Hampshire, amiTexas. Again, the
achievement results for students tested on an annual basis are not cor-
rected for possible bias and should be interpreted with caution.

Table 4

Reading and  Mathematics  Achievement   Results   for  Students Tested  on an Annual
(Fall-to-Fall or   Spring-to-Spring)   Schedule   During  the   1982-83   School    Year

Reading

Mathematics

Grade
	Weighte Number Tested
	d Percentile
		NCfe 2/ Gain Score
	Weighted Number         Percentile
			NCE 2J Gain Score

		Pretest
	Posttest
		Tested
	Pretest
	Posttest
	
2
	70,901
	31
	32
	1-0
	46,036
	38
	42
	1.9

3
	84,606
	25
	30
	2.9
	55,990
	32
	37
	2.7

4
	94,223
	24
	28
	2.6
	63,539
	29
	33
	2.8

5
	96,384
	24
	28
	3.3
	67,646
	28
	35
	4.7

6
	85,966
	24
	29
	3.6
	60,820
	27
	36
	5.2

7
	47,680
	24
	28
	2.3
	31,842
	25
	32
	4.7

8
	43,551
	22
	26
	3.1
	30,598
	26
	32
	3.8

9
	22,277
	22
	26
	2.8
	16,463
	28
	32
	2.3

10
	13,308
	19
	21
	1.6
	8,833
	26
	28
	1.4

11
	9,758
	20
	20
	-0.6
	7,048
	27
	28
	0.9

12
	6,786
	17
	16
	-0.3
	5,044
	26
	27
	0.3

Other Reported Data: States reported data separately for students tested on
a fall-to-spring test cycle and for those tested on an annual cycle. Despite
the substantial bias that can be introduced by fall-to-spring testing, the
majority of students  about two-thirds in school year 1982-83  were
tested fall-to-spring.

Fall-to-spring achievement testing typically indicates gains substantially
and consistently higher than those indicated by annual testing. Fall-to-
spring testing may overestimate true gains by as many as five to six NCEs
in the lower grades. The reason for this disparity is as follows; school
districts typically test their Chapter 1 participants in September, at the
beginning of the new school year, while national standardized test norms are
usually based on October or November testing. Early testing results in over-
estimating true achievement gains because fall-to-spring test results do not
take into account students' rapid, natural achievement growth at the beginning
of each new school year. Annual testing, by contrast, minimizes the likeli-
hood of such overestimates.

Table 5  illustrates the problem  of overestimation of achievement gains that
typically results  when   students are tested  on a  fall-to-spring test  cycle.

reported   by   States   and  local   school   districts.

The data   in  Table   5   weren of